International Journal of Leading Research Publication

E-ISSN: 2582-8010   •   Impact Factor: 9.56

A Widely Indexed Open Access Peer Reviewed Multidisciplinary Monthly Scholarly International Journal

Call for Paper Volume 7 Issue 4 April 2026 Submit your research before last 3 days of to publish your research paper in the issue of April.
Submit Research Paper Join as a Reviewer

Plagiarism is checked by the leading plagiarism checker TurnItIn

Call for Paper

Volume 7 Issue 4
April 2026

Submit your research paper

Indexing Partners
Academia Advanced Sciences Index Bielefeld Academic Search Engine CiteSeer DRJI Google Scholar Independent Search Engine & Directory Network ISI (International Scientific Indexing) Issuu Mendeley Research Networks
RefSeek ResearcherId - Thomson Reuters ResearchGate Scirus Scribd Semantic Scholar UTeM - Universiti Teknikal Malaysia Melaka Wiki for Call for Papers WorldCat Zenodo

Reducing Hallucination and Improving Relevance in Telugu LLMs Through Prompt Design

Author(s) Dr. SriSudha Garugu, Ms. Harshitha Dudala, Mr. Jayanth Reddy Mandadi, Mr. Sai Kiran Ladineni
Country India
Abstract Telugu, a classical Dravidian language spoken by over 83 million people, remains significantly underrepresented in the training data of modern large language models. This imbalance causes two interrelated problems: the models frequently generate plausible-sounding but factually wrong outputs — a phenomenon called hallucination — and their responses often miss the point of what a Telugu user is actually asking. This paper tackles both problems head-on through a combination of LoRA-based fine-tuning and structured prompt engineering, without
requiring expensive full-model retraining. We built a purpose-made bilingual instruction dataset covering eight distinct cross-lingual task variants, fine-tuned Mistral-7B-v0.3 on it, and then applied four complementary prompt design strategies: role declaration, confidence gating, chain-of-thought reasoning, and dialect-script cues. The results speak for themselves - accuracy jumped from 56.25% to 87.5% and precision from 54.55% to 90.0%, confirming that
thoughtful prompt design is one of the most practical tools a developer has for improving Telugu NLP quality right now.
Keywords Semantic Search, FAISS, OpenAI Embeddings, Vector Database, Natural Language Processing, Similarity Search, Information Retrieval, Artificial Intelligence, Machine Learning, Intelligent Search System
Field Computer > Artificial Intelligence / Simulation / Virtual Reality
Published In Volume 7, Issue 4, April 2026
Published On 2026-04-17
View / Download PDF File

Share this



CrossRef

CrossRef DOI is assigned to each research paper published in our journal.

IJLRP DOI prefix is
10.70528/IJLRP

Downloads
Research Paper Format Copyright Permission Form and Undertaking Form Cover Page Vol 6 Isu 12 Cover Page Vol 6 Isu 10 Cover Page Vol 6 Isu 9

All research papers published on this website are licensed under Creative Commons Attribution-ShareAlike 4.0 International License, and all rights belong to their respective authors/researchers.

CC-BY-SA Open Access

About IJLRP
Fees & Payment
Current Issue
Publication Archive 		Submit Research Paper
Track Submission Status
Publication Guidelines
Publication Ethics
Peer Review & Plagiarism 		Join as a Reviewer
Editors & Reviewers
Reviewer Referral Program
Get Reviewer Membership Certi. 		Website/Journal Policies
Usage Policy
Content Policies
Privacy Policy
Contact Us +91-9687-828-838 editor@ijlrp.com